We consider a manufacturer that offers two types of prioritized warranties for its product. Type 1 warranty guarantees a shorter turnaround time than type 2 warranty. Hence items covered by type 1 warranty receive higher priority in repair service. When an item under warranty fails, the manufacturer sends it to one of several repair vendors for repair, who are under contracts to provide repair service for the manufacturer. The manufacturer pays each vendor a fixed fee per repair assignment. While an item is at the vendor under or awaiting repair, a linear holding cost is incurred by the vendor and a linear good-will cost is incurred by the manufacturer. We first study the admission control problem for a single vendor that can either accept or reject an incoming repair assignment in order to maximize its own profit. We analyze the optimal control policies under three criteria: individual optimization, class optimization, and social optimization. By exploiting two proof methods, value iteration algorithm and sample path analysis, we prove that the optimal policy under each criterion has switching-curve structure. We also compare the optimal policies under the three criteria mentioned above and show that (i) the class-optimal policy accepts more high priority customers but fewer low priority customers than the socially optimal policy, which has interesting socioeconomic connotation, (ii) the individually optimal policy accepts more high priority customers than the class-optimal policy, while it can accept either more or fewer low-priority customers than either of the other two optimal policies. We then consider the warranty repair allocation problem which the manufacturer faces. The manufacturer's goal is to allocate the repair work in such a way that the total cost (including fixed cost and good-will cost) is minimized. The complexity of the problem makes the attempt to find the optimal policy very unlikely to succeed. Therefore, we turn our attention to heuristic routing procedures. We develop an effective and robust index-based policy by applying a single policy improvement step to a well-chosen static routing policy. We evaluate the index-based policy and compare it with other heuristics via simulation.